Machine learning characterization of sperm quality for sperm selection for assisted reproduction technology

Abstract

A user computing entity provides a specimen score for use in selecting a sperm for use in a fertilization event. The user computing entity provides the score by obtaining specimen image data comprising imaging data of one or more specimen, the imaging data corresponds to at least one type of imaging; generating a specimen scoring request including the specimen image data; providing the specimen scoring request for receipt by a network computing entity; receiving a specimen response comprising a respective specimen score for the one or more specimen, the respective specimen score for the one or more specimen generated by a machine-learning trained specimen analysis model; processing the respective specimen score for the one or more specimen to generate a graphical representation of the respective specimen score; and causing display of the graphical representation of the respective specimen score.

That which is claimed: 
     
       1. A method comprising:
 obtaining, by a user computing entity, respective specimen data comprising imaging data, the imaging data comprising at least one of respective Raman micro-spectroscopy data or respective quantitative phase imaging (QPI) data for each individual specimen of one or more specimen, the imaging data for each individual specimen of the one or more specimen captured via non-invasive imaging of the individual specimen, wherein the imaging data is captured of the one or more specimen with the one or more specimen contained within a slide that spatially isolates individual specimen of the one or more specimen, and wherein the respective specimen data is associated with sample data that provides at least one of demographic information or medical history information of a source of the individual specimen; 
 generating, by the user computing entity, a specimen scoring request comprising at least a portion of the respective specimen data; 
 providing, by the user computing entity, the specimen scoring request for receipt by at least one network computing entity; 
 receiving, by the user computing entity, a specimen response comprising a respective specimen score for each of the one or more specimen, the respective specimen score for each of the one or more specimen generated by a machine-learning trained specimen analysis model executed by the at least one network computing entity and trained to receive a representation of the imaging data comprising the at least one of respective Raman micro-spectroscopy data or respective QPI data and a representation of the sample data as input and, as output, provide a prediction of a fertilization event outcome for the individual specimen, the prediction determined based at least in part on DNA characteristics of the individual specimen ascertained from at least one the representation of the imaging data or the representation of the sample data, the specimen response provided by the at least one network computing entity; 
 processing, by the user computing entity, the respective specimen score for at least one of the one or more specimen to generate a graphical representation of the respective specimen score; and 
 causing, by the user computing entity, display of the graphical representation of the respective specimen score. 
 
     
     
       2. The method of  claim 1 , wherein the respective specimen score provides an outcome prediction for an assisted reproductive technology process performed using the respective specimen of the one or more specimen and the respective specimen is usable for a fertilization event. 
     
     
       3. The method of  claim 1 , further comprising receiving the sample data providing source information for the one or more specimen, wherein the specimen scoring request comprises the sample data and the specimen score is generated by the machine-learning trained specimen analysis model based at least in part on the sample data. 
     
     
       4. The method of  claim 1 , wherein the machine-learning trained specimen analysis model is configured to analyze a feature vector corresponding to the respective specimen of the one or more specimen and generated based at least in part on the imaging data of the respective specimen. 
     
     
       5. The method of  claim 1 , wherein the machine-learning trained specimen analysis model was trained using a self-supervised machine learning technique and each of the one or more specimen is an individual sperm. 
     
     
       6. The method of  claim 1 , wherein the respective specimen data comprises imaging data corresponding to both vibrational microspectroscopy and quantitative phase imaging. 
     
     
       7. The method of  claim 1 , wherein the specimen analysis model is trained to determine a probability that the individual specimen includes a particular sex chromosome. 
     
     
       8. The method of  claim 1 , wherein the machine-learning trained specimen analysis model was trained based on training image data labeled with information corresponding to at least one of successful fertilization, embryo development, or live birth out comes of a corresponding specimen shown in the training image data. 
     
     
       9. The method of  claim 1 , wherein the one or more specimen are held in isolation in the respective wells of the slide while the user computing entity obtains the respective specimen data, generates the specimen scoring request, provides the specimen scoring request, receives the specimen response, processes the respective specimen score, and causes display of the graphical representation of the respective specimen score. 
     
     
       10. The method of  claim 1 , wherein a phase map is generated based on the respective QPI data and provided as input to the machine-learning trained specimen analysis model as at least part of a two-dimensional or three-dimensional feature vector determined based at least in part on the specimen data. 
     
     
       11. A user computing entity comprising at least one processor and a memory storing computer-executable instructions, the computer-executable instructions configured to, when executed by the at least one processor, cause the user computing entity to perform the steps of:
 obtaining respective specimen data comprising imaging data, the imaging data comprising at least one of respective Raman micro-spectroscopy data or respective quantitative phase imaging (QPI) data for each individual specimen of one or more specimen, the imaging data for each individual specimen of the one or more specimen captured via non-invasive imaging of the individual specimen, wherein the imaging data is captured of the one or more specimen with the one or more specimen contained within a slide that spatially isolates individual specimen of the one or more specimen, and wherein the respective specimen data is associated with sample data that provides at least one of demographic information or medical history information of a source of the individual specimen; 
 generating a specimen scoring request comprising at least a portion of the respective specimen data; 
 providing the specimen scoring request for receipt by at least one network computing entity; 
 receiving a specimen response comprising a respective specimen score for each of the one or more specimen, the respective specimen score for each of the one or more specimen generated by a machine-learning trained specimen analysis model executed by the at least one network computing entity and trained to receive a representation of the imaging data comprising the at least one of respective Raman micro-spectroscopy data or respective QPI data and a representation of the sample data as input and, as output, provide a prediction of a predict fertilization event outcome for the individual specimen, the prediction determined based at least in part on DNA characteristics of the individual specimen ascertained from at least one the representation of the imaging data or the representation of the sample data, the specimen response provided by the at least one network computing entity; 
 processing the respective specimen score for at least one of the one or more specimen to generate a graphical representation of the respective specimen score; and 
 causing display of the graphical representation of the respective specimen score. 
 
     
     
       12. The user computing entity of  claim 11 , wherein the respective specimen score provides an outcome prediction for an assisted reproductive technology process performed using the respective specimen of the one or more specimen and the respective specimen is usable for a fertilization event. 
     
     
       13. The user computing entity of  claim 11 , wherein the computer-executable instructions are further configured to, when executed by the at least one processor, cause the user computing entity to perform the steps of receiving the sample data providing source information for the one or more specimen, wherein the specimen scoring request comprises the sample data and the specimen score is generated by the machine-learning trained specimen analysis model based at least in part on the sample data. 
     
     
       14. The user computing entity of  claim 11 , wherein the machine-learning trained specimen analysis model is configured to analyze a feature vector corresponding to an individual specimen of the one or more specimen and generated based at least in part on the Raman micro-spectroscopy data of the respective specimen. 
     
     
       15. The user computing entity of  claim 11 , wherein the machine-learning trained specimen analysis model was trained using a self-supervised machine learning technique. 
     
     
       16. The user computing entity of  claim 11 , wherein each of the one or more specimen is an individual sperm. 
     
     
       17. The user computing entity of  claim 11 , wherein the machine-learning trained specimen analysis model was trained based on training image data labeled with information corresponding to at least one of successful fertilization, embryo development, or live birth out comes of a corresponding specimen shown in the training image data. 
     
     
       18. A network computing entity comprising at least one processor and a memory storing computer-executable instructions, the memory and the computer-executable instructions configured, when executed by the at least one processor, to cause the network computing entity to:
 receive specimen data, wherein the specimen data comprises specimen imaging data comprising at least one of a representation of respective Raman micro-spectroscopy data or a representation of respective quantitative phase imaging (QPI) data for each individual specimen of one or more specimen, the imaging data for a respective specimen of the one or more specimen captured via non-invasive imaging of the respective specimen, wherein the one or more specimen are held in isolation in respective wells of a slide while the imaging data is captured, and wherein the specimen data is associated with sample data that provides at least one of demographic information or medical history information of a source of the individual specimen; 
 execute a machine-learning trained specimen analysis model to process at least a portion of the specimen imaging data and the sample data to generate a respective specimen score for at least one of the one or more specimen, wherein the machine-learning trained specimen analysis model has been trained to predict fertilization event outcomes for individual specimen based at least in part on DNA characteristics of the individual specimen determined based at least in part on the at least one of the representation of respective Raman micro-spectroscopy data or the representation of QPI data and the sample data; and 
 provide a specimen response comprising the respective specimen score for receipt by a user computing entity. 
 
     
     
       19. The network computing entity of  claim 18 , wherein the memory and the computer-executable instructions are further configured, when executed by the at least one processor, to cause the network computing entity to, prior to generating the respective specimen score, train the machine-learning trained specimen analysis model by training a specimen analysis model, the specimen analysis model trained using a plurality of training imaging data instances where each training imaging data instance (a) comprises at least one of vibrational microspectroscopy data or QPI data for a single sperm and (b) is labelled with an outcome of a fertilization event including the single sperm. 
     
     
       20. The network computing entity of  claim 18 , wherein the memory and the computer-executable instructions are further configured, when executed by the at least one processor, to cause the network computing entity to, prior to generating the respective specimen score, train the machine-learning trained specimen analysis model by:
 pre-training the specimen analysis model using initial training data comprising at least one collected training data instances or generated training data instances; 
 identifying additional training data instances, the additional training data instances comprising imaging data representing one or more morphological features of specimen indicated by a notification generated responsive to the pre-training of the specimen analysis model; and 
 further training the specimen analysis model using the additional training data instances.